Bolt.new revenue: a dev-tools company's AI pivot to $4M ARR
StackBlitz had under $1M ARR before launching Bolt.new, an in-browser AI app builder. Bolt.new revenue reached $4M ARR in about four weeks.
StackBlitz was founded in 2018 and spent years building WebContainers, a way to run Node.js inside the browser. It was impressive technology attached to a small business: under $1M in ARR (founder-reported, The New Stack). Then a new model made the old technology suddenly valuable. About four weeks after launching Bolt.new, the company was reported at $4M ARR (media-reported, Evil Martians).
Why did StackBlitz build Bolt.new?
When Claude 3.5 Sonnet arrived in mid-2024, the team saw that a strong coding model plus a full dev environment in the browser could build and run whole apps from a prompt. The pain it addresses is setup: building an app normally means installing tools, configuring environments and deploying. Bolt removes that.
How does Bolt.new work?
A user describes an app. Claude 3.5 Sonnet generates code, which runs immediately in WebContainers, a WebAssembly-based in-browser runtime. Netlify provides one-click deploys and Supabase provides databases. The whole loop runs in a browser tab, so there is nothing to install and the app the model writes is running in seconds, which is what makes the demos so persuasive.
How was Bolt.new built?
They built Bolt.new quickly on top of existing tech: WebContainers, Claude 3.5 Sonnet for code generation, Ruby on Rails for accounts and billing, Netlify for deploys and Supabase for databases. Team size is not disclosed with a dated primary source.
How did Bolt.new market and price itself?
It was announced at ViteConf in October 2024. A conference keynote, demos that spread on X, a Supabase tutorial video and a fast-growing user subreddit followed.
Pricing is token-based monthly plans; the team added top-ups after heavy users kept running out of tokens.
What challenges did Bolt.new face?
Usage-based pricing created the first friction: heavy users kept running out of tokens, so the team added top-ups. Rapid growth also meant scaling a product built in weeks. Shipping fast left little time to polish. Competition from Lovable and Base44 arrived in the same months.
How much revenue does Bolt.new make?
Bolt.new reached $4M ARR about four weeks after launch, in November 2024, which implies about $333K a month (media-reported, Evil Martians). By December 2024 it had 100K+ weekly users (media-reported, Evil Martians). Higher later figures come from podcast summaries and are not used. Total revenue is not disclosed.
The timeline: StackBlitz founded in 2018; the team tests Claude 3.5 Sonnet with WebContainers in mid-2024; Bolt.new launches at ViteConf in October 2024; $4M ARR in November 2024.
Why did Bolt.new work?
Years of deep tech can suddenly become valuable when models catch up. WebContainers gave Bolt something competitors lacked: a full runtime in the browser.
The team pivoted fast when a new capability fit its stack. And a ready audience of developers at ViteConf gave it a launch stage.
What can founders learn from Bolt.new?
Years of deep tech can suddenly become valuable when models catch up.
Pivot fast when a new capability fits your stack.
Usage-based pricing needs easy top-ups. When your most engaged users hit a limit, make it easy for them to pay more instead of leaving.
Frequently asked questions
How much revenue does Bolt.new make?
$4M ARR about four weeks after launch, in November 2024 (media-reported), about $333K a month. Total revenue is not disclosed.
What was StackBlitz's revenue before Bolt.new?
Under $1M ARR (founder-reported, The New Stack).
What powers Bolt.new?
WebContainers (an in-browser runtime), Claude 3.5 Sonnet for code generation, Ruby on Rails for accounts and billing, Netlify for deploys and Supabase for databases.
How many users does Bolt.new have?
100K+ weekly users as of December 2024 (media-reported, Evil Martians).
Sources
- bolt.new from StackBlitz: how they surfed the AI wave – Evil Martians, 2024-12
- Interview with Eric Simons – The New Stack